Transcription
All right, continue through the summer and end a uh a quiet month of July for the S&P. A brutal market for factors, for tech, for the AI trades, and for hedge funds.
Um, I'll go through all of it, try to give you uh both some experiential perspective from having been through uh a bunch of what situational awareness uh went through uh starting early in my career. So hopefully everything I'm going through will help to some degree.
I'm going to talk a lot about is that the end of what went on and what to look for. I do think there's a high probability that this was a uh a cleansing event. uh I don't see everything just racing back up, but I do think that when you hit those types of events, and I'll go through the reasons why, uh it's probably more important than you realize just because of what happens when a hedge fund actually goes under of that size.
Uh number eight, compute demand is insatiable. I just want to make sure that if I'm wrong about everything that you guys watch on a weekly basis, number eight's the most important thing because I say it on every interview. I say it every week here. Compute demand is insatiable. There is no way for the supply side to keep up with the demand of Q compute currently and there are no indications that that's case. But most of the bearish rhetoric comes from the fact that people are worried that we're overbuilding. We're not going to get the revenues. I'm going to cover all of that this week based on the earnings, based on podcasts again at a time when people are so negative on AI from so many different directions. And as someone who puts their um neck out on the line believing and using and talking about this, the sentiment is very easy for me to hear based on the emails I'm getting from subscribers.
Um, yes, people are losing money. I've lost money this month in in in a big way and I my positions are reduced. So, I get the fact that when uh positions go against you, it's tough, but that's the market and that's what goes on. Uh, I'm not here to tell you everything is a guarantee. Everything I do is based on probabilities. The one thing that I do think is an absolute guarantee is that the market structure has changed forever and I'm going to cover that as well. And I think it's important particularly for institutions uh hedge funds, mutual funds.
Let's get into it. So this is something that I ran. Now this was through uh Wednesday. So this is the Goldman Sachs H or VIP hedge fund index. So they create an index of the most crowded positions or the most owned positions in the hedge fund world based on their data. I created took that index and the GS hedge is relative to the S&P 500. This is a monthly look back since the index was created which goes all the way back to 2001. Down 12% for the month. So that would mean that in a month that the S&P finishes unchanged. Now again, this is through Wednesday. The S&P was down about 2% through Wednesday. Um, this was by far the worst relative performance of that crowded hedge fund positioning, which again, hedge funds are smart. They're doing their homework. If it's crowded position, that means there's got to be good fundamentals to it. Uh, worse than during the month of Lehman. That's what that month is there, too. and but but aside from that doubled any other month. So this was a massive shock to the hedge fund side in terms of crowded positions.
I did a rolling four day. Seven of the 12% occurred in the four days from Friday of last week until Wednesday of this week. So Friday, Monday, Tuesday, Wednesday. That four-day rate of change on that was the worst ever. And that includes during COVID when everything was collapsing at the market level. Again, markets going down. This is happening in a tape where the S&P is not moving and the VIX is low. This will get into the market structure side.
So, we started getting these stories. This actually came out on Thursday morning. Hedge funds face demands to stump up collaterals. AL stocks tumble. So, at this point, news was starting to circulate that there was a problem. So this is Thursday morning before the market. This is in the FT. So this is during European time. And clearly something is happening. And then you in the article it says Goldman Sachs, JP Morgan were among the banks that had asked clients to stump up additional collateral. According to several people familiar with the matter, the additional collateral were automatically triggered by market volatility. Goldman said in a recent note that the buildup in gross leverage in the first five months of the year was the largest cumulative increase it have recorded since it began tracking the data in 2016. And this was already at all-time highs according to data which I'll show you guys later.
In a sign of the pain for hedge funds long short hedge fund strategies were down over 1% multistrats were down over 1.7. U basically the last time all these strategies were down more than 1% a single day was during COVID. However, and this is probably the most important line in the first thing I want you to think about, which I did in my midweek market update this week for subscribers, which is hedge fund strategies were still on average up more than 10% for the year. Now, I don't think this was through Wednesday. I think this is through the Friday before. As I told you, it was a brutal week, but then you got the bounce back. So, figure they're down. They're up less than 10, but that's still up a lot. And I talk about this often. During these situations, chip stack matters. So when your chip stack is still positive, you're a lot more likely to just hedge your stuff up and then come back to fight another day.
Now the situational awareness Leupold [clears throat] seeks to raise capitals after the AI route. So again, this is in the FT and we find out on Thursday at 11:30ish that Citadel buys situational awareness stock portfolio after big losses in AI. So there's a lot to talk about in this. Um I just want to bring this up that this is the kind of thing that does mark bottoms normally. [clears throat] Do I think there's a lot of other funds that are possibly in a similar situation? Yes, there probably are. The difference is number one, this fund is very large. And number two, as I go through this, I want you to just pay attention. Citadel is obviously much larger. Situational awareness is reported to have had 45 billion worth of positions. Um, I'll go through some of the details, but these are the articles or two separate articles that I just want to make sure you hear because based on my experience, this is the reason why I think there's a high probability that this was a market cleansing event.
Even if there's say another hundred hedge funds that are in the same situation as situational awareness, but the question is, if you take a hundred hedge funds of a billion dollars, you go through what they've hedged up already, does it really matter that much? And I'm going to say no. Um the hedge fund is working with prime brokers including Bank of America, Goldman Sachs, JP Morgan to meet margin calls or wind down positions in orderly way. Key thing in this multiple prime brokers know what's going on. The banks had been marketing. It's a key word marketing a portfolio of hedge funds long and short holdings before the market opened on Thursday.
I was involved in LTCM in a very close manner both before uh in the six months before during the closeown I took over the S&P book when it wasn't even completed yet. Um I've seen this from LTCM back but I was also very involved in the GARP quant unwind of 2007 before the great financial crisis and also Archos. Um there's been other mini ones but those are the main ones. I don't consider Melvin Capital to be in the same side, but what you're dealing with is once the marketing of the portfolio begins and they start looking for buyers. Basically, you're showing people po people other people's poker hands and they're going going to go out there and they're going to hedge their book or front it or do whatever.
In another one, multi strategy hedge fund firm Millennium Management also had a bid. So when you combine Millennium and Citadel, these firms both combine trillions of dollars of of positions. So of course, if they've seen what's going on and to the extent, there's a variety of things that are going to happen, but the reality is this fall that we saw here, which is really those last four days. And the reason I have this chart up, this is Bloom Energy relative to Adobe. And effectively when you go through their positions, the biggest ones they have, the short positions that are rumored, this is a pretty good proxy for their book. The reason I bring this up is they go around shopping things because they've given back a sizable portion. And when you give back a sizable portion of your gains, even if they are up 80% year to date, with half or threequarters of that being their private anthropic position, you're still getting very close to the fact where you can't meet cash because the anthropic is not going to go through. So they have to find someone to buy their book and give them the cash necessary to meet the collateral calls. So you get this massive sell-off and then immediately at that day you've got a change that's gone on. You've had a cleansing event.
So in my mind given the fact that Bloom Energy gave back everything that it had to the point where we got the last 13F holdings, which is right here, which is where we saw situational awareness's positions. I do think that you have to start from the perspective of is the unwind over. It's definitely not over from people that need to get out. They never are. But the question is, are there more people in a position now that have already hedged their portfolio? Especially when you're dealing with the prime brokers and the largest banks out there who have been the banks were already out there asking for more collateral. And then you have the two biggest hedge funds or two of the biggest hedge funds in the world also seeing the situation and having positions on internally as well.
So I did this during the week and I just want to bring this up. This was an especially important risk management week with the Federal Reserve major hyperscaler earnings and the end of a brutal month for momentum. That's what made this situation worse. And honestly, this is what obviously took Leupold over the edge is that going into the week from last Friday when there was clearly a problem. You had a big week coming up where the Fed and I'm not a tinfoil person, but the fact that Citadel came out with a news story saying surprise rate hike uh was their call for Kevin Worsh. and you have the hyperscaler earnings which could cut capex and that rumor was flying around and you've already had a brutal month which means people are playing with a short a short stack. That's why I went through this to kind of go through midweek. So for those of you who are subscribers I've done a midweek I think every week this week in July or at least the last three weeks. Um and the reason is because there was a lot going on in this but also a lot going on in the thematic portfolio which uh I've obviously put on the the website. It's part of this because it's correlated to everything that's gone on.
So this was the Citadel security seeing Worsh delivering surprise rate hike. So when you put it all together, you're back with I saw LTCM. I saw the Quan unwind in 2007. Was involved in it. Had positions on in both of those by Archos. I was not involved in it other than having a variety of 30 some odd portfolio managers reporting to me and all of them end up in the same situation where you get the point and then margin calls from several global investment banks. They all know what's going on. Block trade started to happen in this case. It's a little bit of all of them to me. I had said before that the market movements were reminding me more of LTCM than anything and the quant unwind as well. LTCM is different in the fact that that leverage was impacting the world and all of the markets around the globe except for the US markets were already falling in this case almost all the markets around the world are above their 200 day moving average and continuing to move higher.
So, one of the elements that came on again last week that was an indication is someone and it was either Leopold or preparing for while they're going out to go out and figure out how to market this. There were multiple billion dollars of SMH puts that were bought uh to hedge things. So, SMH falls violently in the four days and guess what? You have this massive volume spike basically on the lows. Um I I you know for everyone who's looking for things in here and again this is actually an awesome looking chart for the size of the unwind that's happened. I said this before this is an ABC at this point. Um the lows to me are critical and I'll go through that as well. Uh, but I mentioned Micron last week. There's a neckline of a head and shoulders formation right here that we broke below and that we've broken back above.
Now, at this point, this was the most interesting part. Uh, the day before the news articles came out, uh, this was the second day in a row that uh people had called me up saying that someone was being unwound and something big was happening. So, this is the day after Kevin Worsh. So, this is Wednesday. The market sold off. The S&P was down 120 points, which it made all back the day, the next day. This was clearly the day that the positions were taken on hold by I'm assuming Citadel and they take these positions on and then there's a massive futures position that goes on and the reason that you can see this is this is the total year-to- date volume of the E- Minis the S&P electronic futures from 2:00 to 4:00 so into the close except for one day which was June 4th which happened to be the gap down here which was a big volume day. This was the biggest volume for E- mini futures in that 2hour period into the close of the entire year. So that's how you know that a lot of the selling was done in a pukage basis by just hedging what had gone on.
So when someone comes to the market and let's assume that the reports are correct and he sold about 15 to 20 billion dollars of positions at the market bid down a certain amount. Well, the futures, it's not surprising they get knocked down one one and change percent. Uh, they probably bought everything down as an index 3% down, 4% down, 5% down. So, I used to do this when I traded the S&P book. Um, these types of things go on and you make markets on them and that's what happens and I think that's what went on.
Now, it doesn't get past the fact that as of Thursday morning before we got the bounce, this is where the momentum factor was for technology. Down 40% year to date. Worse than anything during the dotcom bubble by almost 20% except for this one period here I that one month there. So again, historic month for industrial momentum again before the market opened on Thursday, down 32%. I mean double any other month. This one's the most important, especially for you multistrats, but also for um anyone who's factor uh focused. If you strip out the factor risk of your portfolio, then you're running a factor neutral portfolio that's also going to be sector neutral. This is sector neutralized momentum as of again Thursday morning down 21%. Um now again when you look at the months that were up these were unprecedented up months. This is an unprecedented down month. So again we're giving back gains and this is the way it goes. You stretch the rubber band too far one way you get levered. This is the reason why I did the AI midcycle slowdown because we had stretched the rubber band too much one way. I got crap for getting out of Micron way too early and then I got crap last week for saying I was buying Micron below 900. Guys, I'm going to go through this as it goes on. When you see these types of moves, if you think you're going to be able to pick tops and bottoms, I've never in my lifetime be able to pick tops and bottoms. What I'm trying to do is increase positions based on a basian philosophy of the news that's coming for me. If you're watching this to try and t pick tops and bottoms, especially in the speed chess world, it's not going to happen. This down here is already extended. It was extended down here. It was extended here. You're trying to stay solvent by waiting until the panic subsides. And we saw a panic situation that historically signals the end.
In terms of those four days, this is more shocking. This is the 4-day rate of change again into the close on Wednesday for sector neutralized momentum. So again, this is a long short basket down 17% that is sector neutral. So this is momentum across all sectors. It just shows the extent of the damage.
Now I've talked about this before and this is why everything has changed going forward. Um, this is as of uh the close on Friday. This is the current 30-day and 60-day realized V of the Morgan Stanley tech momentum factor. There is no way that this doesn't keep the risk at lower levels. So, even if people have derisked, they're not going to re-risk anywhere near where they were before. [clears throat] Partly because of the draw down, partly because some of them are going to be in the hole. But a lot of it has to do with until this volatility comes down and I don't expect it to come down quickly and I still believe structurally it's going to be higher than it's been in the past. This is the main problem. This is something which is beyond belief to be up at 110 when the prior highs were around 70 which means a normal move right now in the market taking the square root of 252 108 divided around divided by the square root of 252 which is 16 gives you about 6 to 7% a day and that's market or you know that's that's market neutral in the fact it's dollar long dollar short these types of moves are huge.
Now, if you take it a step further, what I did is I took all of the market neutralized factors. So, forget just the tech momentum side, which is tech, and I took it for beta, I took it for quality, I took it for momentum, and I took it for value. And I just went through and gave it an equal weight. So, divide each of them by four. They're uh 60-day volatility. And here's where we are. 38. We're above. So this gets into any kind of optimized portfolio where you're risk weighting even your factors. Th this is something that has been trending higher for some time. Just look, you don't see anything where this stuff trends higher for a long period of time. This is the impact of AI and it is never going away. And I'm going to go through the reasons why.
We Here's more um on the liquidations that went on and just some more of the stats. Hedge funds just sold more tech than any point in Goldman's 10-year record. momentum factor volatility at its highest in 45 years outside a recession. So I don't even have data going back 45 years. Gross and net exposure is off five and four-year highs now in the middle of the pack. Six weeks of that and hedge funds are still up 9% through June. Again through June, one of the best first halves in 20 years. So again, that was through June.
So I'm going to remind you guys, I wrote this paper back in May. Bubbles, parabas, and speed crashes. This is exactly what we saw. So I wrote this before the peak. We're going to see more bubbles. We're going to see more par parabas. And what comes with that are speed crashes. So it's going to look like a bubble. Basically, we're compressing time. And the reason we're compressing time is because AI is compressing time. The model capabilities are going so fast and we're entering into a world of AI agents which are on 24/7. It's the end of human beings. Now, this has been happening anyway. If you guys have been down in the New York Stock Exchange anytime uh since the 1990s, it looks a lot different today than it did back then. We've already been in this situation. Machine learning has already been a part of our world. It's just that now it is ubiquitous and everyone has it. You have to incorporate that into what's going on now.
We not only saw the liquidations for hedge funds in Korea, we've basically been in a 3-w week unwind of positions there. So again, we've cleansed positions, but you're not going to be able to take that kind of size again. So don't expect this to go back to the same direction because what chased it, as someone who focuses a lot on Bitcoin and on uh crypto, you need the energy of retail, and we're just not going to get back to that anytime soon. I talked about that with silver, gold, and Bitcoin. The difference is with the AI trades, for those of you, I say it every week, the majority of positions are above the 200 day moving average still. So of the AI positions, I think 74% are still above and 84% of my basket are still the 200 day moving average is still going higher. So we're not in any position where any damage was done.
Hedge fund leverage, this was as of the third quarter of 2025 and this is data from the Securities and Exchange Commissions form PF. You can see how much it had gone up since co hedge funds had been having performance. this has gone higher. It's not just the basis trade. It is also equities. So, this is not just fixed income. It is across a range of places.
Now, I want to bring this up because I wrote this paper on the art of unlearning the Fed, which we'll get back to, but I just want to read this line because this is what goes for everyone who's a risk manager. If you're out of fun and your risk manager doesn't watch my videos, I would give them this because this is a structural change. AI is progressing at an unprecedented pace and in a world defined by exponential innovation, relying on backward-looking models is too slow. This was me writing about what Scott Besson had said in an interview on the All-In Pod when he was going through what was going to have to happen with the Fed.
Here's the way that it's doing it. First of all, AI is compressing economic time. The model capabilities are improving rapidly. So, what used to take 10 years in innovation is now taking a year. We went through this in the first quarter. Terminal value harder to estimate. You don't know if you're going to be in business in three years if you're built on code. Think Adobe. Think Salesforce. People still have no idea. So there was multiple compression because we were taking off terminal value. All this feeds into the factors. If all of a sudden we don't know if a company is going to go, they can go down 40% without their earnings changing. This is not good for risk management in terms of having this instability. Hedge funds operate on a much shorter clock than the assets they own. This is going to be a big deal, especially as we get into tokenization where things are going to trade 24 hours a day. Are the hedge funds going to be in a position to trade this stuff 24 hours a day? How much more miserable is life going to get if you're down big over the weekend? The more that markets are open or trading, the more likely you are to make an emotional response that is bad. Leverage turns valuation uncertainty into liquidity risk. That is important because when you get falling confidence in terminal value, it raises volatility, tightens financial conditions. We're seeing all of these play out. So, I don't think the prime brokers are going to allow leverage to go to the levels they have because I don't think the volatility that we saw is coming down.
Ubiquitous AI risk tools create a false sense of confidence. This is like watching people drive down the street with all-wheel drive when it's icy and snowy out, thinking that they're protected. Everyone right now has AI risk tools. You can just say optimize this portfolio. What is a way to reduce my risk, you're just at a point where there's a bubble in optimization and thinking that your portfolio is hedged because it's using, as Scott Bessant alluded to, the way that you come up with an optimized portfolio is that you're using past correlations and past volatility. As I've shown you in the factor stuff, this is no longer relevant unless you believe mean reversions happening. And I'm telling you, the market structure has changed forever because of AI. It is never going back. Tokenization is here. Terminal value is gone forever. All of this stuff is happening. You're seeing things like companies that are private like Anthropic grow so fast, the fastest in the history of companies, but also seeing companies go out of business faster than they've ever. The greater terminal value uncertainty leads to multiple compression. That's what we're seeing. So, the S&P 500 earnings are growing at over 20% a year, but the multiples compressing because the stocks are not going up to match it.
If you don't believe the AI agents are here, Robin Hood keeps releasing things. Here's an AI trading agent now reads charts, RSI, MACD, and 16 more. This means retail has access to the same tools that every quant hedge fund has. Now, we're going to get through the changes that are happening and how disastrous this is, especially for the month of July because I am sure there are lots of people. I get so many people that reach out saying they've been buying S&P puts, they've been buying NASDAQ puts, this is going to crash. I'm not kidding you when I say that very smart people who I respect keep constantly calling up with this stuff.
Here's what's gone on. This is tech momentum. the one-year realized V at 67 and a half. Here is the NDX realized V for the same time period. Here is the relationship over time. This is insanity. You're at one of the lowest levels of realized volume in the NDX and you are parabolic on the volatility. This is where RAAS, mutual funds, everyone who doesn't have the same time horizon as a hedge fund should be having a conversation with me. We're getting those conversations. This is about market structure. You can benefit significantly from this. And from the people that I talked to regularly, they did very well reducing risk back in May, even though it was early, because they're in a position right now to start building up some of the risk on fundamentals that they like. There's going to be more speed crashes, guys, than there ever have been. If you don't believe that chart, here is that factor neutral V that I said, which is at 36, which again takes an equal weight of quality beta momentum and value sector neutral factors equal weight relative to the six contract of the VIX. So for everyone who's bought S&P vault or sixmonth, this thing tends to go higher and we could be in an event right now where multiple hedge funds are going under.
If you ask me as I'll get through, the only two areas when I've spoken to the risk group at the Fed that I'm worried about, I'm worried about insurance companies and I'm worried about hedge funds. Those are the only groups that I think have the leverage capable of this going up here. That would be a credit event. I think that needs to involve either a hedge funds, but since Citadel Millennium are up big for the year and they're involved in this stuff and they're the best risk managers in the world, we're probably in a situation where this is more likely to decompress over time. And that's my baseline. If you ask me what's more likely to happen, I don't think there's going to be a major event. I think there's going to be multiple compression and I think there's going to be a lot of disappointment, but I think deleveraging is going to happen for a long time because of this staying up here. So, if you ask my gut, I don't think we're mean reverting anymore. We've already seen a structural higher level here than it was all back here. And this is not a short time period. This basically stayed between five and 15 from 2000 to 2020. Since then, it's been from 15 to 30. We're in a structurally higher volatility, so you're just going to have to deal with it.
Here's another one. This is taking the VIX which is the index implied V and this is taking the VIX EQ which is the equity V. So equity V has broken away. This is the longterm relationship between the two and we've got a huge difference here and effectively it started right around Chat GPT. So, I just think that equity vault because you can go out of business much very quickly is going to stay at the high side and index vault is going to stay on the lower side relative to it. Um, again, if you want to reach out, especially if you're an RAIA where I'm having these conversations, you can do it weekly, you can do it monthly, but I think being on top of the factor side, being on top of the AI from a bigger perspective, asking the questions, that's where the alpha is going to be generated. So, you can reach out on there.
Um, this is a bull market still. 67 new highs in the S&P yesterday, the most since February. So, as this whole thing has been going on in July, we've got that. We got 72% of the S&P 500, greater than our 50-day, 200 day, 68% above the 20-day. I mean, again, everything's been fine during this. Here's where we finished the month, unchanged. four sectors down, seven up. This is not normal compared to what went on in LTCM and what went on during the quant unwind. It is similar to Archos. Argos ended up going higher from there. But I think you just have to pay attention.
Here's a chart on the S&P and remember that was the day where we sold off at the end of the day and we've immediately not only gone back up. The next day we went higher and now we closed way above it. I'm going to get into some of the technicals on that, but just be aware that every time we kind of push this ball underneath the water, it pops out. Here's equal weight S&P, the monthly returns. Look at how many months we've been up with only one month down, which was when the invasion or the bombing of of Iran occurred and oil spiked higher. This is a broad market in terms of the rally that's going on.
Here's the NDX. So we hit the lowest RSI in the last year. You look at what happened the last time we got down to this level and we had a similar type day where we were down and the next day we gapped higher down gapped higher. We'll see what happens. Uh I'm going to go through some of the technicals but remember this is very correlated with my thematic portfolio. the AI trade in general, semiconductors, the NDX did go down and NDX over the S&P actually had its worst month since the.com bubble. Believe it or not, guys, uh worst month since the.com bubble.
Here's SMH. As I mentioned, we had this reverse head and shoulders. I'm telling you, um, again, these lows need to hold, but reverse head and shoulders where you break the neckline, everyone gets short. I'm sure a lot of people got short here, and then we gap higher and then we sit there. Technically, this is what I'd be focused on. If you guys aren't, you know, uh, aware of follow-through days, a follow-through day is confirmation that an oversold bounce may be turning into a real investable rally rather than just a short covering pop. So, the reason I like this and the reason I used it, made a lot of money off this after bottoms. Um, I preferred to find bottoms than tops because generally the sentiments there, you get these very noticeable signals. I was talking about all the things that bothered me beginning in May and the market kept going higher. Situational awareness going out of business and all the things that I showed you in the beginning of this that is very indicative of a cleansing event. Typically technicians look for it a few days after the low often four day uh day four through 10 of an of rally attempt. A major index rises meaningfully. It's common volume is higher blah blah blah. So basically what you're looking for is give it time.
Here's the low. The next day we have the biggest volume since June. This is on the cues. So we get this selloff here. This is the day again that we get the panic from Liupold. So the positions trade war speaks. We get this massive selloff. The next day we trade on higher volume. We close above the highs of that day. Structurally a good thing. Do I think we're probably going to spend time here? They're talking about day four to 10. So, as we move out here, if we hold the lows and if we just kind of build this time in here, that's a good sign. That's what I'm looking for. Um, we'll see what happens.
If you're looking for fundamentals to support this, uh, we're now through 60% of the S&P. Uh, sales growth since earnings growth is a little bit, not a little, it's a lot distorted by uh, anthropic inside this. But when you go through each of these, just look how big these numbers are. It's every sector. More importantly, sales growth 14%. I mean, guys, I keep saying it. It's like it doesn't pay to be to be bearish stocks when this is going on. You have so much fundamental support and revisions as we go through earning season again going higher. So, not only are they beating numbers, not only are they coming through on a historical basis, the revisions keep moving higher. You don't want to be bearish on stocks while this is going on. This is the longest period now in the last five years. It's very difficult. So, we'll see what happens.
Here's my thematic portfolio. So, last week we were over here. I said we're at the 50. It wouldn't surprise me to see us go lower. Blah blah blah. We did. We're back up in here. We're in the 30 some odd. Is that it? I'm like I said, I'm betting on the fact right now with my own portfolio off the lows of that panic low that that was it. If not, there's something bigger going on. Hedge funds could be massively unwinding. We'll see what happens. But that's why we trade these lows. We don't sit there and invest them. Um, I have positions on and things I believe in, which I'll have. And then the positions like Micron and AA ai and uh Marll and some of the other things I've added to. We'll just see where everything goes.
Here's a thematic portfolio on an RSI basis. 14-day RSI hit the same level it did here on the lows here again. We'll see what goes on. But as of right now, I'm leaning towards that. Now, this is the thing that I want you guys to understand in terms of adding. Go back to what I said, how factors are changing and market structures changing. Never before, in my opinion, has looking at the rate of change been more important. Most people I know don't use this. I've always used it since my days in Brazil. Uh I do believe markets move on a rate of change basis and it pays to reduce risk when you get above historic levels. Now in May and into here is when I was reducing my micron position. So beginning around the middle of May a little bit all the time a little bit of Marll until we got up here. Now this is the 63day rate of change. So this is 63 trading days which is three months. We made it all the way up to we were up over 50% on a look back of the prior 63 days for the thematic portfolio. Now over the last 63 days, we're down 1.7%. We have now retraced. We have now done what we needed to do in both price and time. And that's what it takes with rate of change. When I did the consolidation and I showed you the the consolidation patterns that are likely to happen, it's time and price. So we've done that. So, this is just a metric for you guys to use. I provide it every single week in that technical sheet and I'll show you some of the things in there on the 30-day basis. So, a month and a half or a month and a quarter down 15%. So, from as my father taught me, the odds on the AI trade are much better than they were at here. This is on a 30-day basis when I started to reduce micron. So, again, you're at a much better entry point in price and time. I'm not trying to pick the bottoms in this. There's a lot of new news that went on.
Now, for those of you looking to trade the index, which I highly recommend, if you're going to pick names, then trade the names. But if you're going to jump into things, the 10 name concentrated position, I just want to make sure you guys see that it was correlated before. I spent a lot of time to come up with 10 names that are highly correlated to the 100 name index. So, you didn't have to do that. It stayed correlated with it and traded extremely well with it. One of the names in there is Corning. I highlighted this this week on the subscriber video. The thing about corning, it hit the 200 day, it traded up. The reason I have this chart here from December 31st. This day here is when the announcement came from Meta for $6 billion. They had another day here where Nvidia partnered with them. So they do announce a partnership with Nvidia. They announced a partnership with Meta for $6 billion, which according to every LLM that I use is equivalent to somewhere around a hundred million dollar 100 million homes worth of fiber, optical fiber. And the stock got all the way back to where it was basically on the close of that day. This is one of the concentrated names. Now, if you look at it on a rate uh well, I guess I didn't put it in there. I thought I had a rate of change basis on it.
Here are the uh percentage names. So again, I've shown you guys this each week. When you upload the technical sheet, I highly recommend going into co-work and using this to look for this. The reason I highlighted these is July 14th, we had 26% of the names above the 20-day. Here we are now. We've had two weeks where even with all the carnage, it's starting to show signs here of positivity relative to these two weeks. I'm more interested in this one. This is somewhat similar to here. If we start to see this thing move higher again, I think that's a good sign. 84 of the names out of the hundred 200 day slope rising. So, you're looking for signs on that. I'm uploading this to the website. This is combing through the rubble.
So, I went through and I basically wanted to do something and again part of this is to teach you guys how to do this. Um, I highly recommend, again, I do this presentation for RAAS as well in terms of how they can go through in someone's portfolio and be able to show the things that are technically still moving well and be able to do this. So, this is two conditions. I want to make sure that if I'm going through the rubble, number one, I want the uh price above the 200 day moving average. Think Micron, the 200 day moving average is somewhere 500. We're still around 850. And I want the 200 day itself still rising. Those are the names that structurally to me I want to be involved in. That's my favorite technical sign. So I want to find things that have fallen back. And what I do in here is I go through and rank all of them on that. You guys will see the list. It has all the names in it. And so just to give you like a quick side like I mentioned AOI, I ended up buying some this week. And again when I say I bought some, these are all small positions. If you go back to what I had in May, my micron position right now is about 1/5if of the size that it was in May when I was getting out of it. So again, I'm not trying to pick the bottom of this. I'm trying to work into something where I don't think the upside is a good at 800. Do I think it can get to 1,600 over the next year? Yes. At 500, it's a better riskreward. So I'm not convinced yet that we've made the absolute low, but I want to start buying some because I think we're eventually going back up to 1500 to600. Now, in the case of this where the slope is up on the 200 day and the percent above the 200 day is here and we're now 31% down for the month, 27% down over the 3 months verse the 50-day it's below by 34. Like I'm trying to find names that are still above here, still positive here, but are way below in these things. So, I go through the list. I also go through what they've said on their earnings as well to see if the fundamentals still line up.
Here are the two things that I want you guys to start paying more attention to that I have one eye open. So, I'm always as a risk manager looking for something that could change my mind as to what's going on. So, the hyperscaler CDS does not bother me one bit. What does worry me is that the triple C market, the option adjusted spread has this bad looking role to it and this is overlaid with high yield or the junk market. So whenever I see these divergences, the way credit usually goes is the weak start going up and it eventually drags everything in. There is no doubt right now that the amount of capital that's being absorbed by AI is massive and I do believe the disruption happening from AI is real. So, I think we're going to see more and more private credit issues. The question is, is this some of the private credit issues that have been in play really this entire year? Because this started right around the time that the software stuff fell. So, I wouldn't ignore it. And if you guys want to read, I I'm I love good conspiracy theory when it comes to credit. I like to stay on top of things because I do believe that they start with oneoff things and the Dodgers have bothered me as a sports fan spending money in any possible way. And so this whole thing with the Dodgers, Guggenheim, Mark Walter, and private credit, I'm just I would go read this and just pay attention. I don't know what it means. I just know the insurance industry fairly well and I think there's a lot of mispriced assets in there.
Uh Google, I was shocked how many people when I did my video last week told me I was wrong on Google. This was going to collapse. And of course, this was here. Sorry, here we closed below the 200 day. Um I put this in last week. I am shocked how many people don't care about the backlog of Google and tell me it doesn't mean anything. It's not real. I I couldn't disagree more, guys. Uh I'm going to go through a bunch of things on this. I think you guys need to stop reading some of the people who are basically just doing the math like I showed last week and saying these earnings sucked. Clearly, Google's earnings didn't suck. They've already gone back up. I expect fully this company to completely take off and keep going because the numbers are insane in terms of the backlog. And if you didn't believe them, Zuckerberg comments about compute demand from his call. There's just nowhere near enough compute for all the demand. We are getting a large number of offers for the compute that we have. Meta is not building compute ahead of uncertain demand. Zuckerberg said demand already exists. Available compute blah blah blah. You don't believe them because there's stock tradeoff. Sundar Pachai, we continue to be supply constrained. A sign of momentum and rapid adoption. The demand still outpaces the investment. Meta, there's nowhere there's just nowhere near enough compute for all the demand. Microsoft, which traded much higher, customer demand continues to exceed available capacity. That additional inquarter capacity for
Azure was quickly monetized. Can't tell you how many smart people told me that these guys were going to cut capex this quarter. Wrong. Every incremental unit of compute brought online is being absorbed almost immediately while demand continues to outrun supply.
If you don't like those hyperscalers, how about we go to Amazon? Even at that amount, we will still not have enough capacity to meet all of the demand we have in 2026. I believe this dynamic will also be true in 2027, too. Guys, don't be short these things. Honestly, like what are you doing? I I don't like the demand was already we already have for 2028 is striking. This may be the strongest forward-looking comment from the call because Amazon is already seeing substantial commitment several years ahead. The lion share of AWS 2027 compute capacity is already reserved. We have a clear sign of light to strong financial returns. Amazon also explained why free cash flow looks weak before the revenue arrives. It begins spending on data. So again, I want you to go back to what I said about the factor risk. The model capabilities are insane. They are moving faster and faster. This whole open-source thing is stupid. Yes, it will be a part of it. Open router will go through it, but they need to have the cloud and they need to have these companies because they're the only ones with the compute. Nobody else is building the compute. I showed you guys last week that those Chinese companies are so tiny and they're building compute. They have to be run somewhere. So where do you get the chance to use the compute of these Chinese models? It'll have to be because it's on Bedrock or it's on wherever you want, but it has to be on somewhere where there's compute. Somewhere around the line, there has to be hardware that has the compute. There isn't any compute.
Here are the new backlog numbers. 500 billion for Amazon, 500 billion for Google, 700 billion. This is $1.7 trillion between those three of a backlog. Our backlog stands at 496, growing triple digits year-over-year. A large amount of it tends to be deals and agreements that you've signed. They talked about the fact that their new demand is not from OpenAI and Anthropic. The majority of backlog is related to typical GCP contracts for a broad mix of customers. Oh, sorry. This is the one. All sequential commercial RPO growth was driven by commitments from customers outside of frontier model companies. Um, if the people that have scared you are not writing about the facts in the earnings, then what they're saying is not facts. What they're saying is trying to use bear porn on you. You have to at least go through. You can say that these guys know nothing of what they're talking about. That the Mag Seven that are the proven track record winners that their revenue is growing rapidly. They are the ones spending the capex and you're telling me the people writing these garbage newsletters are saying that Google, Amazon, and that they're stupid. There is a risk that it's circular and the fact that Anthropic and OpenAI go out of business, but we'll hit on that next.
Allin podcast. I thought it was a very, very good one. And I give these guys a lot of crap because I think the thing has become way too political, but lately they're back on their uh their game. A lot of good things in there. I'm not going to go read all of it. Um, but they go through the Kimmy K3. They go through this. I thought David Sacks in particular had a ton of useful information on this just because he's connected to the White House and I think he was just very open and honest about all of this stuff. He pushes back against the panic and basically talks about how as much as he hates Anthropic. He talked about how they're the fastest growing tech company in history and that so far their annualized revenue is at 74 billion through the halfway mark of the year. So he's saying they're going to grow 10 times again this year over a hundred billion. And he mentions that OpenAI is now on pace for about 75 billion this year. Guys, that's 175 billion in revenues from two companies that had zero two years ago. How can you be arguing about ROIC when these guys are printing money at that side and the cloud revenue guys are printing $2 trillion during the exact same period? So how people are coming up with this when the companies are getting the demand from the API side and from enterprises at the same time that the com cloud guys are getting it. So it's just misleading completely.
So compute demand versus supply. Sacks goes through it. As far as I'm concerned there is an unlimited appetite for on-demand intelligence. I don't think there's an upper bound for how much intelligence you can tap as long as it continues to get better. I completely agree. He goes through the numbers there on both these things. You have you still have stuff in the labs by Anthropic and OpenAI that is way ahead of the Chinese open-source models. Sam is going to Washington over the next week to talk about GPT 6.0 which is blowing the doors off. Again, the models that are out, these guys have better. If you haven't seen Sam Altman on Invest Like the Best, I highly recommend going through it. I thought it was one of his better performances. Altman says, "OpenAI's conviction to secure massive compute came from seeing an exponential model improvement curve, especially around GPT4. Basically, again, demand would be basically uncapped." He says the biggest near-term compute gains may come from software efficiency. Uh talks about the cyber incident that happened uh with Hugging Face. Uh I'll leave AGI alone and go through it. I summarize this in a little better ways if you guys care about, but um the fear flip because demand exploded. This is the thing I want to make sure. As of October of last year, the probability of these companies being in trouble to me was much much higher than at any time during this year. And the reason was because the the risk was these guys were spending money and the enterprises were never going to adopt it. That human beings were just going to be too slow. The only thing that changed it was that the AI agents came on much faster. And that's the point that a reason I did the portfolio is because the agentic world started and I started putting all these names together way before it went because I had already done the homework. The inference demands were going to be insatiable once AI agents reached. We just didn't know what was going to happen. And it happened at least a year faster. That was how Leopold invested in all these things as well. It just got to that point. What Sam does say is supply is the bottleneck today. Compute bottleneck. The bottleneck has moved over time. He talks about research being a huge one. The amount of demand is just going to accelerate. Personal agents would massively increase demand. This is the Meta side. This is why Meta isn't selling their compute. They could have made investors very happy this week by saying, "We're going to rent out our compute," which they kind of leaked and then they didn't do it. Why didn't they do it? Because if they do it, personal agents are going to massively increase demand. So, I'm sure they've run into this thinking this is coming faster than what we all know and I believe that the oversupply scenario would be models become so smart and efficient that they can do almost anything people need with current compute. Altman believes compute demand is structurally under supply because model quality unlocks new use cases. So basically the Jevons paradox thing gets in. Is there a chance? So what is the most likely scenario for supply to catch up demand based on his views? Software agents squeeze more intelligence out of compute. So this is the efficiency side. Physical world robotics so humanoids eventually expands the entire compute supply chain. Neither of these is happening in the next year. It's just not happening. It it it can't we're not there yet to be able to go through it. And even if we are, the demand side is going to pick up the pace because of the personal agents. That's why I shown you guys the chart of what it means to have personal agents. So agents make compute production dramatically more efficient. Models become so efficient that each task needs less compute. Human attention becomes the bottleneck. So good one to go through. Uh I talked about the revenues. They've seen a major uptick since May. This is the other reason why you have to be positive on these things. Now this makes sense to me and I hate to say it. Where would you expect these to be? These guys are raising so much money. If credit spreads weren't widening and they weren't building in more risk as all of these guys are spending. Of course, there's a risk that any one of them could go out of business if they don't get the revenues in the door. Meta is making a huge bet. They don't have anything now. If they don't actually figure this out, they'll be in trouble. Gavin Baker market is overreacting to hyperscaler credit spreads widening. The fact that spot prices for GPU rentals are at least two times higher than contractor rates is a missing piece. I could read all of this stuff. The reality is at 60 billion per gigawatt that's 1.5 to 2.2 trillion in capex consensus estimates for hyperscaler cash flow 1.3. He's just doing the same point I am which is we don't have enough compute right now guys. Darkh put this out. It I think had over a million views um in terms of X going through it. I highly recommend going through to read it. It's a it's a very contrarian thought. Uh and it really says without going through it in a big way, the part that matters the most. His argument is that expensive compute converts token efficiency into pricing power, widens the frontier premium. So this gets back to the fact that if these guys continue to move at this pace and their intelligent is better but more importantly their token efficiency which comes from the amount of compute they've built. So the moat becomes the compute that they have access to the fact that they can charge again a reasonable price for token efficiency. So to get this to a point where people can understand what it means I had to kind of go through it. Um, and again, Gavin spoke out about this, too. I wanted to give you guys an analogy that could work. Gas and cars. Why the best AI model wins when compute gets expensive? If gas is a dollar a gallon, nobody cares about the fuel efficiency. When it's $10, everyone pays a premium for the car with the best mileage. This is what's going to end up happening is that you might be able to pay cheaper for Kimmy K 3.0, No, but it's not as efficient as the best models. And that situation gets worse and worse because you need more hardware or compute. And if there isn't any hardware or compute to run it, the only place that has it is the place with the compute. If Anthropic has an enormous amount of token efficiency and Disney wants to take a $5 billion movie cost but make it for $100 million, that margin of being able to buy all of the compute that Anthropic has, not only gives them an insane way to make tons of money, but it gets them to destroy their competition as well. So, we're getting to the point where you have to look at the token efficiency. This is the other reason why I've written so much about Vera Rubin. We're getting into a situation where token efficiency is the story, not token maxing. And just remember how much SpaceX was able to do for their compute. So we have a shortage of compute. The models are going faster. They're bringing on more demand because they're getting better and better, but there isn't enough compute out there. So the people that have already built the compute, those prices are going to get higher and higher as SpaceX saw, as Meta saw when they realized that they could sell this stuff out. So just keep it in the back of your mind that we're reaching a point where it's great that there's all these competitive models, but you have to be able to run them on something. And the question is where is that compute going to come from? And if the models are more efficient in terms of the token usage, you're still going to go back to using a a better model except for the stuff that's free that you download on there. So we're not at a point yet where we have the hardware for everyone to be running open-source models. So there's at least another two years in my opinion and that's more than enough time based on what we heard from the hyperscalers this week as to when they're going to get the revenue in the door. It's not a given. I gave you guys a probability last week but it is happening. Now if you still don't believe that the data centers and the compute then just look at DRAM prices. This is the weekly prices of DRAM and again this is the oldest one. This goes back to 2016. This is not the newest memory. They're still going up every week. So, the cyclicality, cyclicality, green, mainly red. Look how many weeks this is going on and it's not stopping. Yes, it's slowing down because the price has gone up so high.
All right, let's bang through the the rest of these. Uh, I did want to do something on the Fed because again, so many people talking about it this week. People telling me some of these strategists were saying it could be 50. So, he holds rates three. The Fed did exactly what Wall Street was expecting. So why are the markets so vexed by Kevin Warsh again? The art of unlearning the Fed. If you didn't read it, go read it. This is the world that I think we're in. Again, I think Kevin Warsh is a smart guy. I think he has talked publicly about the fact that we can't be using the academic approach of the past. We got to have a task force in to figure out the new things to use. The data that we're using now is not only stuff that gets revised. It's old. We need to be more forward-looking. The Fed is not the same Fed and this whole dovish hawkish thing is not there. Now tenure rates are sitting up near 470. Here is again 10-year inflation swap rates. So as of right now since Kevin Warsh went in, we're down here. I don't know how long this will be, but he was saying, "Well, the market did it for us. Real yields are up here. Inflation's come back down. So we can sit and wait." Here are the inflation expectations for the August print around zero.
Dollar positioning. So, this is an interesting one because when the dollar weakens, it's usually a very bullish thing and it's good for gold, it's good for silver. Uh maybe Jackson Hole will have an impact on this. Maybe the dollar yen intervention, but look how big the crowded positioning is on the dollar. It's at the highest level in the last four years. And the dollar has come down somewhat this week. Most of it because of this speculation on the BOJ. And I just want to highlight for those of you trading, let's see, end of April, end of July, end of January. So, this looks like three months, three months regardless. It's not pegged, but it sure seems to be like it's on a Let's just make sure it doesn't go anywhere.
One podcast for you guys to pay attention to to kind of figure out finish out this week. Um, I've heard a lot of people talk about China, uh, and they're not in the same plan on AI and they're looking to just make it free for the rest of the world. This is a great interview. Uh, Ezra Klein interviewed Kevin Rudd, who is the former prime minister of Australia and an expert in Xi Jinping. I loved listening to it. I'm not going to read all this stuff on there, but for those of you trying to understand better uh what China is trying to do and where the balance of power is in terms of AI and how they're viewing it, I think this is a really interesting one. I think Kevin Rudd did a great job. For those of you who care about things, the Marxist Leninist lens combined with the Schumpeterian side, it's really interesting to me. I put this in myself. So, this is me kind of extrapolating what I heard, comparing it to my belief in Joseph Schumpeter. This is the reason why I care so much about Bitcoin. For those of you interested, I think it's a a great um interview worth couple things.
The cyber stuff is getting real. OpenAI and Hugging Face. You had this hacking. You also had a Claude hacking as well. Just worth paying attention to. And the Clarity Act is sitting here. It's still below 40% um depending on which party you listen to. We'll see what happens. Uh again, I don't think this matters in the long term uh for Bitcoin and for crypto. I think it does matter obviously uh for the entrepreneurs that are involved in the space and for the short term, we'll see how it plays out. That's it for this week, guys.
So, final thing, I do believe that we saw something really important this week and that based on my history of dealing with hedge fund unwinds, those lows matter a lot. Um, and how we trade over the next six to seven days, there should be back and forth, but use that follow-through day approach. See you guys next week.